Antifibrillarin Autoantibodies Present in Systemic Sclerosis and Other Connective Tissue Diseases Interact with Similar Epitopes by Kuppuswamy N

Antifibrillarin Autoantibodies Present in Systemic Sclerosis and Other Connective Tissue Diseases Interact with Similar Epitopes By Kuppuswamy N. Kasturi,* Akira Hatakeyama,* Harry Spiera,* and Constantin A. Bona*

From the Departments of *Microbiology and *Medicine, Mount Sinai School of Medicine, New York 10029-6574

Summary Autoantibodies specific against fibrillarin, a 34-kD nucleolar protein associated with U3-snRNP, are present in patients with systemic sclerosis (SSc). To understand the mechanisms involved Downloaded from http://rupress.org/jem/article-pdf/181/3/1027/1106300/1027.pdf by guest on 25 September 2021 in the induction of these autoantibodies, we prepared a series of human fibrillarin recombinant proteins covering the entire molecule and analyzed their interaction with the autoantibodies present in various connective tissue diseases. Our results showed that antifibrillarin autoantibodies are present not only in SSc, as previously reported, but also in a variety of other connective tissue diseases. Patients with SSc (58%), mixed connective tissue diseases (60%), CREST syndrome (calcinosis, Raynaud phenomenon, esophageal dismotility, sclerodactyly, and telangiectasia syndrome) (58%), systemic lupus erythematosus (39%), rheumatoid arthritis (60%), and Sjogern's syndrome (84%) showed presence of antifibrillarin autoantibodies. Results obtained from competitive inhibition radioimmunoassay and Western blot analyses with purified recombinant fusion proteins revealed that these autoantibodies react primarily with epitope(s) present in the NH2- (AA 1-80) and COOH-terminal (AA 276-321) domains of fibriUarin. Autoantibodies reacting with internal regions of fibrillarin are less frequent. Analysis of the hydrophilicity profiles of reactive peptides showed presence of three potential antigenic sites in the NH2- and two in the COOH-terminal regions. While a hexapeptide sequence NH2 terminus of fibrillarin is shared with an Epstein-Barr virus-encoded nuclear antigen, the COOH-terminal region shares sequence homology with P40, the capsid protein encoded by herpes virus type 1. Interestingly, these two regions of fibrillarin also contain the most immunodominant sequences, as predicted by surface probability and the Jameson and Wolf antigenic index. These observations suggest that molecular mimicry might play an important role in the induction of antifibrillarin autoantibodies.

resence of antinuclear antibodies is a hallmark of rheu- (8), PM-Scl (9) and U3 snRNP-associated fibrillarin (10, 11) p matic autoimmune diseases (1). Autoantibodies reacting are targets of scleroderma autoantibodies. Antifibrillarin anti- with nuclear proteins like Smith antigen, (Sm) 1, topoisomer- bodies may characterize a subset of patients with diffuse scle- ase I, centromere proteins, and amino-acyl tRNA synthetase rosis and widespread telangiectasia (12). In contrast to humans, are characteristic for SLE (2), systemic sclerosis (SSc) (3), production of autoantibodies specific for nucleolar antigens CREST syndrome (calcinosis, Raynaud phenomenon, esopha- is seldom associated with the spontaneous development of geal dysmotility, sclerodactyly, and telangiectasia syndrome) systemic autoimmune diseases in animals. However, antifibril- (4), and polymyositis (5), respectively. Although pathoge- larin antibodies can be induced in B10.S and SJL mice after nicity of these antibodies has not been established, they are injection with mercuric chloride (13, 14). useful as markers in the differential diagnosis. In addition to Fibrillarin is a highly conserved 34-kD nucleolar protein the antinuclear autoantibodies, antibodies reacting with specific localized in the fibrillar regions of nudeoli (10, 11). It is as- nucleolar proteins have been reported in systemic autoimmune sociated with U3 snRNP complex, which is thought to play diseases (6). For instance, RNA polymerase I (7), 7-2RNP a role in the processing of preribosomal RNA (15). Fibril- larin has three distinct domains: (a) NH2-terminal domain 1Abbreviations used in thispaper: CREST, calcinosis,Raynaud phenomenon, of 80 amino acids rich in glycine and dimethylarginine; (b) esophageal dismotility, sclerodactyly,and telangiectasiasyndrome; GST, central domai~a of 90 amino acids containing the RNA binding glutathione-S-transferase;ItrIG, isopropyl-/~-D-thiogalactopyronoside;MBP, myelinbasic protein; MCTD, mixed connectivetissue diseases; MS, mul- consensus octomeric sequence; and (c) COOH-terminal a tiple sclerosis;RA, rheumatoidarthritis; Sm, Smith antigen; SS, Sjogern's helical domain of 151 amino acids (16). The genes coding syndrome; SSc, systemic sclerosis. for fibrillarin of Xenopus laevis (17), Physarum polycephalum

1027 J. Exp. Med. The Rockefeller University Press 0022-1007/95/03/1027/10 $2.00 Volume 181 March 1995 1027-1036 (18), yeast (19), and mammalian species (16, 20, 21) have been Texas, San Antonio, TX), rat collagen I (from Dr. R. Holmdahl, cloned and sequenced. The primary structure of fibrillarin University of Uppsala, Uppsala, Sweden), human myelinbasic pro- from these sources shows a high degree of homology to human tein (MBP) (from G. Hashin, Saint Lukes Hospital, New York), fibrillarin. and dsDNA were used as antigens in the direct-binding and com- Antifibrillarin antibodies are currently demonstrable either petitive inhibition RIA. Recombinant fusion proteins containing five contiguous moieties of human fibrillarin were obtained by by immunofluorescence or immunoprecipitation methods. Use cloning the corresponding cDNA sequences into pGEX-2T ~ ex- of these techniques has shown that a small fraction of pression vector (28). scleroderma sera-3.4% (22), 6% (23), and 3.6% (24)- Expression Cloning of Human Fibrillarin cDNA. The coding re- contain antifibrillarin autoantibodies. The major drawback gion of human fihrillarin cDNA (cDNA clone kindly provided by of these methods is their inability to discern the specificity Dr. J. Aris, University of Florida, GainseviUe, FL) and the DNA of the antibodies reacting with large macromolecular com- sequences corresponding to amino acid residues 1-321, 1-85, 86- plexes that were used in such analysis. The nucleolar extracts 148, 149-229, 230-275, and 276-321 of fibrillarin were amplified enriched for U3 RNP complex that were used in immuno- with oligonucleotide primers by PCR (29). The amplified DNA precipitation analyses also contain five other polypeptides with fragments were subcloned into pGEX-2T expression vector, which molecular masses of 74, 59, 30, 13, and 12.5 kD, in addition expresses glutathione-S-transferase(GST). The cloned cDNAs were expressed as GST fusion proteins. Schematic representation of the to fibrillarin (25). Presence of autoantibodies against any one fibrillarin regions that were cloned is shown in Fig. 1. The fol- Downloaded from http://rupress.org/jem/article-pdf/181/3/1027/1106300/1027.pdf by guest on 25 September 2021 of the polypeptides associated with fibrillarin can also bring lowing primers were used for PCR amplification: (a) primer 01: down fibriUarin by immunoprecipitation via the sera that may 5'-atgaagccaggattcagtccc-3';primer 321: 5'-gttcttcaccttggggggtgg- contain these antibodies. Furthermore, fibrillarin is also found 3'; (b) primer 01: 5'-atgaagccaggattcagtccc-3'; primer 85: 5'-aat- associated with other less abundant snRNPs such as US, U13, gtgatggtggagccgcat-3'; (c) primer 86: 5'-attcttccccgactggtttcc-3'; U14, X, and Y (26). In a recent report, out of seven mAbs primer 148: 5'-gattgctgctgctagcttggag-3'; (d) primer 149: 5'-atc- obtained from mercuric chloride treated A.SW mice that ctgggtggtgtggacca-3'; primer 229: 5'-cattgcgatgagcatgcggta-3';(e) precipitated U3 RNP and showed positive nucleolar staining primer 230: 5'-atgctggatctgatctttgctg-3'; primer 275: 5'-ggctga- of Hep2 cells by immunofluorescence, only four mAbs im- ggctgtggagtcaa-Y; and (f) primer 276: 5'-gaggccgtgtttgcctccga- munoprecipitated purified human fibrillarin (27). Y; primer 321: 5'-gttcttcaccttggggggtgg-3'. The 5' and 3' end primers contained BamHI and EcoRI sites, respectively, at the 5' Therefore, we have developed a reliable and sensitive solid- end of the sequence. The amplified DNAs were digested with re- phase RIA, using purified human recombinant fibrillarin pro- striction enzymes, ligated to BamHI-EcoRI-cut pGEX-2T plasmid, teins, first to study the frequency of antifibrillarin autoanti- and the Bb21 competent cells were transformed. The bacteria ex- bodies present in scleroderma patients and then to determine pressing the recombinantphsmids were selected on ampiciUinplates. whether they are present in other connective tissue diseases. The recombinants were grown in I ml Lubria broth (LB) medium Our results demonstrate that antifibrillarin antibodies are and induced with 1 mM isopropylB-D-thiogalactopyronoside (IFIG) present in high frequency not only in SSc but also in a va- for 5 h at 37~ The bacterial lysates were run on SDS-polyacryl- riety of other connective tissue such as diseases like mixed amide gels, and the recombinants expressing appropriate molec- connective tissue disease (MCTD), SLE, CREST syndrome, ular size fusion proteins were selected. Plasmid DNAs were se- rheumatoid arthritis (RA), and Sjogern's syndrome (SS). By quenced by the dideoxy chain termination method (30), with an preparing a panel of six overlapping fibrillarin fusion pro- oligonucleotide primer 5'-gcatggcctttgcaggg-3', corresponding to 855-871 of the pGEX sequence, to confirm that the inserts were teins, we have also determined whether these autoantibodies cloned in frame. are directed against a common epitope(s) or a characteristic

set of epitopes in each disease. Our results showed that the o loo 20o 3?0 ~1 ~=~ autoantibodies from various connective tissue diseases exhibited Number a similar immunochemical specificity suggesting that the same 321 or similar epitopes are recognized. However, in contrast to 85 A autoantibodies present in SSc, MCTD, SLE, and CREST syndrome, which reacted primarily with the epitopes located at 86 148 c ~49 the NH2-terminal region, autoantibodies present in RA and 229 D 230 275 SS reacted equally well with the epitopes present at the iiiiiiii NH2- and COOH-terminal regions. E 276 321 F Materials and Methods.

Sera. Sera were obtained from patients with various connec- Schematic diagram showing regions of tive tissue diseases such as SSc, CREST syndrome, MCTD, RA, human fibrillarin cloned as fusion proteins SS, and SLE and control healthy subjects were stored at -20~ Figure 1. Schematicdiagram showingregions of human fibrillarinex- until use (the patients were classifiedby the criteria described by pressed as fusionproteins. DNA fragmentscoding for definedregions of the American Association of Rheumatism). A total of 237 serum fibrillarin A-F generatedby PCR were cloned into pGEX-2T expression samples (56 SSc, 81 SLE, 22 MCTD, 11 CREST syndrome, 38 vector to produce six in-framefusion proteins as described in Materials RA, 15 SS, and 14 heahhy subjects) were used in this study. and Methods. FibriUarinproteins containing 13 hydrophilicregions were Antigens. Recombinant human fibrillarin fusion proteins, hu- kept intact to preservethe potentialantigenic sites formedby linear amino man Sm antigen (kindly provided by Dr. H. Dang, University of acid sequences.

1028 FibrillarinEpitopes Recognizedby Autoantibodies Antigen-binding Studies. Direct-binding assay. Microtiter plates Total bacterial lysates and purified recombinant proteins run were coated with recombinant fusion proteins or GST (as control) on SDS-PAGE and stained with Coomassie blue R stain are at 5 #g/ml in carbonate buffer, pH 9.0, at 4~ overnight. After shown in Fig. 2. The level of induction of recombinant fu- washing, the plates were blocked with 3% BSA in PBS for 1 h sion proteins varied depending on the insert sequence. The at room temperature, washed three times, and incubated with 1:400 expected molecular mass of fibrillarin fusion proteins FP-A, dilution of sera for 2 h. After thorough washing, 5 x 104 cpm FP-B, FP-C, FP-D, FP-E, and FP-F were 60.0, 34.5, 32.5, of 12SI-labeled rabbit F(ab')2 anti-human y antibody in 50 #1 was added and incubated for a further 2 h. After washing, the bound 34.1, 30.6, and 30.6 kD, respectively. The observed sizes of radioactivity was measured in a gamma counter. Cpm bound to fusion proteins correspond well to this (Fig. 2). All recom- GST coated plates were subtracted from the cpm bound to fibril- binant proteins were full-length products. Only FP-C fusion latin fusion protein-coated plates and expressed as specific cpm protein showed lower molecular mass derivatives, indicating bound. proteolytic cleavage in specific regions of fibrillarin. Sequence Competitive Inhibition Assay. 10/~g/ml of serum IgG, purified analysis of recombinant plasmid DNAs confirmed that the by use of protein A-Sepharose or 1 #g/ml of affinity-purified insert DNAs were inframe with the gent coding for GST antifibrillarin antibodies, was incubated with various amounts of protein. collagen I, Sin, dsDNA, MBP, or fibrillarin in 50 #1 for 1 h at Frequency of Antifibrillarin Antibodies in Sera of Patients with room temperature, and then transferred to fibrillarin-coatedplates Connective Tissue Diseases. The presence of antifibrillarin an-

following the procedure described previously (31). The cpm bound Downloaded from http://rupress.org/jem/article-pdf/181/3/1027/1106300/1027.pdf by guest on 25 September 2021 to fibrillarin was measured as above. The inhibition was expressed tibodies in the sera of individuals with various connective as a percentage. tissue diseases was studied by solid-phase RIA with purified Western Blotting. Purifiedfibrillarin fusion proteins were run fibrillarin fusion protein FP-A. Background binding to GST on 10% polyacrylamidegels in the presence of SDS (32). Separated was determined by parallel assays on GST-coated plates. Cpm proteins were transferred to nitrocellulose filters (33) at 100 V for bound to GST protein was subtracted from total cpm bound 4 h at 4~ blocked with 5% BSA in PBS, and then incubated to fibriUarin fusion protein. The specific binding activities for 1.5 h with a 1:500 dilution of human sera or 0.1 #g/ml of mouse of sera from patients with various connective tissue diseases anti-human fibrillarin mAb in PBS containing 5% BSA and 0.05% are presented as cpm bound in a scatter diagram (Fig. 3). Tween 20. After washing with PBS containing Tween 20, filters The sera showing three- to fourfold higher binding than that were incubated for 1.5 h at room temperature with 125I-rabbit of the control healthy subjects were considered positive. To P(ab')2 anti-human Igy or ml-labeled rat anti-mouse &K mAbs confirm the results obtained by RIA and exclude any false at 2 • 105 cpm/ml. Filters were washed, dried, and autoradio- positives, we tested all available positive sera on Western blots graphed as previously described (31). IndirectImmunofluorescence. NIH 3T3 cells were grown on glass containing purified fibrillarin fusion protein and GST. The coverslips in 35-mm dishes. After 2 d in culture at 60% confluence, sera were preabsorbed with GST-coupled Sepharose 4B to re- the cells were washed with PBS and then fixed with 2% parafor- move the antibodies that might react with GST. We also in- maldehyde in PBS, pH 7.4, for 15 rain at room temperature. The cluded a mouse antifibrillarin mAb and four sera from healthy fixed cells were permeabilized by exposure to 0.2% Triton X-100 humans as positive and negative controls, respectively. Rep- in PBS for 5 rain on ice. After permeabilization, cells were rinsed resentative Western blots are shown in Fig. 4. The results three times for 10 rain with PBS and incubated with 50/zg/ml showed a strong positive correlation between ILIA and Western of IgG antibodies or 5 #g/ml of affinity-purifiedantibodies diluted blot assay methods. The high sensitivity of the immunoblot in 1% goat serum for 1 h at room temperature. The unbound anti- analysis method is clearly demonstrated by the strong fibril- body was removed by rinsing three times for 10 rain with PBS latin band signal in the blot incubated with mouse antifibril- and then incubated with affinity-purifiedgoat F(ab')z anti-human IgG antibody conjugated with FITC at a dilution of 20 #g/ml latin mAb at 0.1 /zg/ml. for 1 h. The unbound second antibody was removed by washing The data summarized in the histogram (Fig. 5) show that with PBS, and the cells were mounted in 90% glycerol, 10% PBS '~58% of SSc patients have significant titers of antibodies (pH 8.0) containing 4% n-propyl gallate (34). The cells were ex- that bind to fibrillarin fusion proteins. A similar percentage amined under fluorescence illumination or laser light in a micro- of patients with CREST syndrome (58%), MCTD (60%), scope (Carl Zeiss, Inc., Thornwood, NY). RA (59%), and SLE (39%) also exhibited the presence of these antibodies. To our surprise, >84% of patients with St Results showed high titers of antifibriUarin antibodies. Thus, our data Expression and Purification of Recombinant Fibrillarin Pro- clearly show that antifibrillarin autoantibodies are present at teins. To identify the immunodominant B cell epitopes of high frequency not only in patients with SSc but also a va- fibrillarin, the PCK-amplified DNA fragments coding for riety of other connective tissue diseases. various moieties of fibrillarin (Fig. 1), each containing one Epitope Specificityof Antifibrillarin Antibodies. KIA was per- to three hydrophilic regions, were cloned into pGEX-2T ex- formed by use of fibrillarin fusion proteins FP-B, FP-C, FP-D, pression plasmid, Thus, a panel of six in-frame contiguous FP-E, and FP-F to determine the domains of fibrillarin that fibrillatin fusion proteins were generated as described in interacted with the autoantibodies. The sera exhibiting sig- Materials and Methods. Bacterial cultures expressing recom- nificant titers of antifibrillarin antibodies were further tested binant proteins were induced with IPTG, and the fusion pro- by direct binding assay with purified fibriUarin fusion proteins were purified by affinity chromatography on a gluta- teins FP-B to FP-E Out of 56 SSc sera, 31 showed significant thione-S-Sepharose gel column as previously described (28). binding to fibrillarin fusion protein FP-A, 30 to FP-B, 5 to

1029 Katsuriet al. Figure 2. Analysis of recombinant fibrillatin fusion proteins. Coomassie blue- stained polyacrylamide gels show lysates of bacteria expressing fusion proteins A-F (A) and the purified fusion proteins (B). Fu- sion proteins were purified by affinity chro-

matography on glutathione~S-Sepharose gel Downloaded from http://rupress.org/jem/article-pdf/181/3/1027/1106300/1027.pdf by guest on 25 September 2021 as described in Materials and Methods. The observed molecular sizes of fusion proteins A-F correspond well to the predicted molecular masses of 60.0, 34.5, 32.5, 34.1, 30.6, and 30.6 kD, respectively.

FP-D, and 8 to FP-F (Table 1). Only 1 sample failed to bind showed a high frequency of binding to the COOH-terminal to any fusion protein other than the complete fibrillarin mol- portion corresponding to amino acid residues 276-321 (FP-F). ecule. The titer of antibodies against individual epitopes present This is comparable to the binding activity to the immuno- in this sera may be low and is not quite sufficient to show dominant epitope present on FP-B. This suggests that anti- above the background when individual fusion proteins were FP-F antibody may be an additional marker to discriminate used, or it is a conformational site that was lost as individ- ILA and SS from other connective tissue disorders. The binding ual fusion proteins were prepared. Sera from MCTD, SLE, specificity and the pattern of reactivity of autoantibodies were CREST, ILA, and SS showed high frequency of binding ac- confirmed by Western blot analysis. Samples of positive sera, tivity to the fusion protein, FP-B, containing the NH2-ter- selected randomly, were tested at 1:500 dilution on Western minal region. Comparison of binding profiles of sera from blots. The binding profiles of several sera were examined in SSc, MCTD, SLE, and CILEST disease groups revealed no Western blots containing fusion proteins FP-A to FP-F. Rep- specific patterns of antibody reactivity. The frequency of an- resentative Western blots are shown in Fig. 6. Distinct pat- tibody reacting with the immunodominant epitope present terns of binding by different sera demonstrate that a number on FP-B and FP-F did not show significant differencebetween of epitopes are recognized by the autoantibodies, confirming the above disease groups. However, sera from RA and SS the results obtained by direct-binding ILIA. This assay also shows that the antibodies demonstrated are specific and do not react with GST. The amount of antibodies bound to the 6000 fusion protein FP-B band in Western blots appears to be some- 5000 what lower than that detectable in ILIA (data not shown).

4000 This may be due to reduced availability of some conformational epitopes present on FP-B after SDS denaturation. 3000 I 9 o +o Specificity of Antifibrillarin Antibodies. To determine whether I r ~ ~ .+ 2000 the autoantibodies present in these disease groups are specific 9 I,, ~ to fibrillarin, we tested the binding activity of autoantibodies 1000 +.+.+ by competitive inhibition ILIA by use of several autoantigens 0 as competitive inhibitors (collagen I, Sin, dsDNA, and MBP are target autoantigens in SSc, SLE, and multiple sclerosis

z [MS]). Results of inhibition ILIA of four antibodies are shown in Fig. 7. The specificity of affinity-purified antibodies ob- Figure 3. Binding activity of sera from various connective tissue dis- tained from a SSc serum that was also used in Western blot eases to fibrillarin. The binding activity was measured by RIA with sera analysis is shown in Fig. 7 A. Preincubation of purified anti- at 1:400 dilution as described in Materials and Methods. Optimum dilu- bodies with chromatographically purified soluble recombinant tion of the serum to use in the assays was determined by preliminary assays. Cpm bound to GST was subtracted from the cpm bound to fibrillarin fibrillarin inhibited the binding of autoantibodies to fibrillarin- fusion protein. The mean values of triplicate assays are shown in the scatter coated plates, while other autoantigens used as inhibitors did diagram. Each dot represents an individual serum. not show any significant inhibition. Results of inhibition assays

1030 Fibrillarin Epitopes Recognized by Autoantibodies Fig,,re 4. Demonstrationof the presence of antifibrillarin autoantibodies in connectivetissue diseases. Purifiedrecombinant human fibrillarin and Downloaded from http://rupress.org/jem/article-pdf/181/3/1027/1106300/1027.pdf by guest on 25 September 2021 GST were separated on 10% SDS-PAGE,blotted to nitrocellulose, and reacted with sera as described in Materials and Methods. Human sera were used at 1:500dilution. Mouseanti-human fibrillarinmAb (0.10/~g/ml) and sera from heathyhuman subjects wereused as positiveand negativecontrols. Representative autoradiograms of Western blots reacted with control sera, SSc (PSS), MCTD, SLE, CREST, RA, and SS are shown. of sera from SSc, MCTD, and SLE shown in Fig. 7, B-D, to the polyclonal nature of the antibodies present. In other further demonstrate that the autoantibodies detected by RIA words, in these sera, antibodies that are specific for other are specific to fibriUarin and are devoid ofpolyreactivity. Similar nucleolar proteins are probably more abundant than antifibril- results were obtained with sera from patients with RA and larin antibodies. Staining of fibrillar regions of the nucleolus SS (data not shown). was more evident when affinity-purified antibodies were used. Indirect ImmunofluorescenceAnalysis. Total serum IgG from Corrfocal microscopy confirmed that the staining of the fibriUar SSc patients, and affinity-purified antifibrillarin antibodies, and granular regions of the nucleolus is characteristic of from the fusion protein FP-A affinity column were used for antifibrillarin antibodies (Fig. 8 G). determining the pattern of binding to nuclear components on freshly cultured 3T3 cells by immunofluorescence and con- focal microscopy. The number of nucleoli per cell vary de- Discussion pending on the cell type used. The 3T3 fibroblasts show two Autoantibodies reacting with nucleolar proteins are pro- to several nucleoli (10). IgG from SSc patients showed bright duced in connective tissue diseases such as SSc, SLE, and SS nucleolar staining with very weak or no staining of nucleo- (6). Autoantibodies present in SSc show diverse binding pat- plasm (Fig. 8, B). On the other hand, affinity-purified anti- terns in immunofluorescence staining, suggesting that a va- fibrillarin antibodies, in addition to brightly staining nucleolar riety of autoantigens might be recognized. The molecular material, also stain the nucleoplasm feebly but consistently. nature of some of these autoantigens has been described re- The weaker staining of nucleoplasm by total IgG may be due cently. A distinct nucleolar protein, fibrillarin, associated with U3-snRNP, has been reported as a target autoantigen recognized by autoantibodies present in diffuse scleroderma (10). Immunofluorescence staining with antifibrillarin antibodies 100- shows staining of fibrillar and granular regions of the nucleoli 8O

60 Table 1. Epitope Specificity of Anti-Fibrillarin Autoantibodies 40 n Disease Fibrillarin FP-B FP-C FP-D FP-E FP-F 20 PSS 31/56 30 0 5 0 8 MCTD 13/22 10 1 6 0 2

0 RA 22/38 14 0 6 0 17 SS 13/15 8 0 3 0 11 Figure 5. Frequenciesof patients with antifibrilllarinantibody activity CREST 6/11 4 0 2 0 4 in connectivetissue diseases. The numbers of sera exhibiting two- to fourfold higherbinding activitythan the mean valueof sera from healthyindi- SLE 30/80 22 0 5 0 10 viduals are shown as percentages in the histogram.

1031 Katsuriet al. (9). The distribution of fibriUarin staining parallels RNA polymerase I staining (7). Our results clearly demonstrate that affinity-purified antifibrillarin antibodies stain fibrillar regions of the nucleolus (Fig. 8). Since the immunofluorescence method does not allow precise identifcation of the antigens involved, we used a sensitive RIA with recombinant fibrillarin fusion proteins to define B cell epitopes. In this paper, we present data demonstrating that antifibrillarin antibodies are produced not only in SSc, as described previously (12), but also in a variety of other rheumatic diseases. Thus, the frequency of antifibrillarin antibodies occurring in MCTD, CREST syndrome, and RA is comparable to that observed in SSc, while these antibodies are more prevalent in SS (84%) but less common in SLE patients (39%). There is a striking difference between our results and those previously reported with respect to the frequency of antifibrillarin autoantibodies in scleroderma. While the results Downloaded from http://rupress.org/jem/article-pdf/181/3/1027/1106300/1027.pdf by guest on 25 September 2021 obtained by immunofluorescence and immunoprecipitation methods showed an incidence of 3.4-6% (22-24), our results obtained by RIA showed a frequency of 58% among scleroderma patients. Positive sera identified by RIA were also confirmed to be positive in Western blot analysis. High correlation between RIA and Western blot analysis results rules out the possibility of high false positives in RIA. The dis- crepancy between our results obtained by RIA and those obtained by immunofluorescence and immunoprecipitation methods can be explained by (a) higher sensitivity of ILIA compared with immunofluorescence and immunoprecipitation Figure 6. Western blot analysis of autoimmune sera from connective tissue diseases. Selected examples of sera reactive with fibrillarin fusion methods, and (b) use of purified recombinant human fibriUarin proteins are shown. Western blots containing recombinant fusion pro- in R.IA in which all epitopes are available for interaction with teins A-F were incubated with a 1:500 dilution of autoimmune sera and antibodies. Thus, use of RIA and Western blot analysis with developed with lzsI-labeled rabbit F(ab')2 antibody specific for human Ig~ purified recombinant human fibrillarin allowed the detection chain. (A) SSc serum. (B) MCTD serum. (C) SSc serum. (D) SLE serum.

100 100- A C 80 80-

=e6o ~60- f = .=,40 ~40-

20 20- ~ .... _=/ 5 10 15 20 i= 5 10 15 20 Antigen concentration ug/ml Antigen concentration uglml Figure 7. Demonstration of the specificity of antifibriUarin autoantibodies by competitive inhibition 100 100 ILIA. Various concentrations of an- B D A Fibrillarin = DNA tigen, 1-20 /~g/ml, were mixed 8O 80 9 Sm with 1/~g/ml of affinity purified an- D ColII tibody or 10 #g/ml of IgG from 6O 6O A MBP g different sera, preincubated, and then transferred to fibrilhrin-coated ~40 "~.40 plates, and the inhibition was cal- _e culated as described in Materials and 20 2O Methods. (A) Affinity-purified antibody isolated from an SSc serum. 0 0 5 10 15 20 (B) SSc serum IgG. (C) MCTD Antigen concentration ug/ml Antigen concentration ug/ml serum IgG. (D) SLE serum IgG.

1032 Fibrillarin Epitopes Recognized by Autoantibodies Downloaded from http://rupress.org/jem/article-pdf/181/3/1027/1106300/1027.pdf by guest on 25 September 2021

Figure 8. Indirect immunofluorescencestaining of 3T3 cells with antifibriUarin antibodies. Freshly grown 3T3 cells were fixed with paraformalde- hyde, incubated with SSc serum Ig or affinity-purified antifibrillarin antibodies, and stained with FITC-labeled F(ab')2 goat anti-human IgG antibodies. Phase (A, C, and E) and immunofluorescencepatterns (B,/9, and F) are shown. (,4 and B) Cells incubated with SSc lgG. (C and D) Cells incubated with af~nity-purified antifibriUarin antibodies isolated from a SSc patient. (E and F) Cells incubated with FITC-labeled second antibody alone. (G) C~nf~ca~micr~sc~peimage~fthe~bri~arregi~ns~fnuc~e~iince~sstainedwitha~nity~uri~edanti~bri~arinantib~dies. xlOO(A-EandF); x250(G).

of antifibrillarin autoantibodies in the sera of patients with fibrillarin, and the fusion protein FP-B, which contains the other connective tissue diseases. It is of interest that the pres- NH2-terminal moiety. About 15-30% of SSc patients also ence of antifibriUarin antibodies has been reported in certain had antibodies that bound to the COOH-terminal region malignant diseases such as hepatocellular carcinoma (35). of fibrillarin, FP-F, while a small percentage of sera also bound Antifibrillarin antibodies present in connective tissue dis- to FP-D, the internal helical region. The autoantibodies that eases are highly specific to fibrillarin antigen. Results from bound to the NH2-terminal portion are distinct from those competitive inhibition RIA demonstrate that binding of that bind to the epitopes present in the COOH-terminal re- affinity-purified antifibrillarin antibodies and autoantibodies gion because all the sera that showed binding to FP-B did present in the sera to fibriUarin antigen-coated plates is in- not bind to FP-F. Western blot analysis confirmed R.IA results. hibited only by soluble fibrillarin and not by other autoan- In Western blots, the antibodies present in some sera recog- tigens such as DNA, Sin, collagen type III, and MBP, which nized only NH2-terminal fusion protein, while others rec- are target antigens in SLE, SSc, and MS. These data, as well ognized only COOH-terminal peptide (Fig. 6). Antibody as low binding to BSA or GST, suggest that antifibrillarin profiles obtained for RA and SS, by use of recombinant fu- autoantibodies are not endowed with polyspecific properties. sion proteins, showed higher correlation of the presence of The variation in the degree of inhibition between different anti-FP-F antibody with the disease. The variation in the epi- sera might be due to differences in the concentration, affinity, tope reactivity and specificity of autoantibodies may be and epitope specifcity of antibodies present in different sera. influenced by factors such as the duration or severity of the The data obtained from RIA and immunoblotting of fibril- disease, HLA haplotype of the patient, and nature of the larin fusion proteins, encompassing the entire fibrillarin poly- therapy. peptide sequence, show that the immunodominant B cell epi- Swiss-Protein data base search revealed the presence of ho- topes are located in the NH2-terminal glycine-arginine-rich mology between the NH2-terminal portion of fibrillarin and region, since most of the SSc sera, as well as sera from MCTD, R.NA binding proteins from various sources and EBV-encoded SLE, and CREST syndrome, react with FP-A, the complete nuclear protein (Fig. 9). The NH2-terminal portion of fibril-

1033 Katsuriet al. 69 80 to be determined. The second possibility is that environmental Fibrillarin NRGP~RGGKRGN :::::: factors, chemicals, or drugs might play a role in presentation EBV encoded NP SGGRGRGGSGGR of these self-peptides to T lymphocytes, leading to their am- 334 345 plification, and thereby help in the activation of autoreactive 311 321 B cells. The presence of elevated levels of antifibrillarin anti- Fibrillarin GVYRPPPKVEN .:::::: bodies in mercuric chloride-treated mice supports this possi- RSV(type i) P40 AVYRPPPHSAP bility (13, 14). The third possibility is molecular mimicry, 533 544 in which the autoantibody response is induced in the host Figure 9. Comparison of amino acid sequences of fibrillarin peptide by infectious agents. The dassical example is the cross-reactivity motifs sharing homology with foreign antigens. Hydrophilic fibrillarin of anti-Streptococcus A antibodies with cardiac myosin (39). peptide motifs (potential antigenic sites) exhibiting sequence similarity with a nuclear protein encoded by EBV genome and capsid protein P40 There are increasing examples showing presence of shared encoded by HSV type 1 are shown. :, amino acid identity;., conserved antigenic peptide sequences between viral, bacterial, parasitic, changes. and autoantigens (40). Sharing of peptide sequence homology between MBP and viral proteins (41), adenovirus-12 E1B protein and gliadin A (42), gp210 protein reacting with autoantibodies from patients with primary biliary cirrhosis (43), larin (fusion protein FP-B) contains three hydrophilic regions leucine zipper region of C-myc and V-myc with pT0 Ku au- Downloaded from http://rupress.org/jem/article-pdf/181/3/1027/1106300/1027.pdf by guest on 25 September 2021 corresponding to aa residues 17-28, 33--48, and 66-84. Based toantigen (44), KAI.,1 antigen of river blindness pathogen on the analysis of surface probability and Jameson and Wolf Onchocera votvulus and 60-kD KO/SS-A autoantigen (45), antigenic index of these regions, as predicted by GCG anal- regions of topoisomerase I with CMV (26) and mammalian ysis (Genetic Computer Group, Madison, WI), the aa res- type C virus P30~8 (46) are well documented. Sharing hex- idue sequence 66-84 appears to be the most irnmunodominant apeptide sequence homologies between fibrillarin and EBV- one in this region. This fusion protein is recognized by a encoded nucleoprotein and HSV type 1-encoded P40 sug- majority of autoimmune sera. Interestingly, this region con- gests that molecular mimicry might play a significant role rains a hexapeptide sequence, GKGKGG, which is shared with in the induction of autoimmunity and production of autoan- EBV-encoded nuclear antigen (36). The next most commonly tibodies reacting with fibriliarin. recognized fusion protein, FP-F, by autoimmune sera, con- The data presented here have twofold importance. First tains the COOH-terminal portion of fibrillarin. This region is the demonstration that autoantibodies specific for fibril- contains two hydrophilic regions at aa residues 281-306 and larin epitopes are produced in a number of connective tissue 314-321. The first hydrophilic region bears high homology diseases in addition to SSc. Second is the demonstration of with yeast fibrillarin as well as some self-proteins such as the presence of autoantibodies against immunodominant epi- chicken myosin heavy chain and microtubule associated protein topes located at the NH2- and COOH-terminal regions of 1A (37). The most antigenic peptide in this region is aa se- fibrillarin. In addition, presence of homology between the quence 311-319, GVYtLPPPK (Fig. 9), bearing sequence ho- NH2-terminal region of fibrillarin and the nuclear protein mology with capsid protein P40 encoded by HSV type 1 (38). encoded by EBV on the one hand and capsid protein P40 One may ask how antifibriUarin autoantibodies are induced encoded by HSV and the COOH-terminal region of fibril- in patients with various connective tissue diseases. The first larin on the other hand suggests that common latent viruses possibility is that the peptides that were generated from the might play a role in the induction of autoantibodies. Fur- nucleolar protein fibrillarin, derived from the endogenous thermore, correlation between the high incidence of antifibril- pathway, are transported to the surface, where they are rec- larin antibodies and the presence of EBV infection in SS pa- ognized by fibrillarin-specific lymphocytes. In this mecha- tients supports this. Studies are in progress to isolate mAbs nism, it remains unclear why antifibrillarin antibodies are pro- against various fibriUarin epitopes and determine whether the duced only in patients with connective tissue diseases and antibodies specific for EBV-encoded nuclear protein or P40 not in healthy subjects or in patients afflicted with other dis- interact with fibrillarin peptides. Such experiments might shed eases. Whether MHC or analogous molecules play a role in light on the possible role of an antigen mimicry mechanism presentation of the peptides derived from fibrillarin remains in the production of autoantibodies against autoantigens.

We wish to thank Dr. John P. Aris for kindly providing us with human fibrillarin cDNA clone. This work was supported by program grant 5P01AI24671 from the National Institute of Allergy and Infectious Diseases. Address correspondence to Dr. C. A. Bona, Department of Microbiology, The Mount Sinai School of Medicine, Box 1124, One Gustave Levy Place, New York, NY 10029-6574.

Received for publication I August 1994 and in revisedform 31 October 1994.

1034 Fibrillarin Epitopes Recognized by Autoantibodies References 1. Tan, E.M. 1984. Autoantibodies to nuclear antigens: their im- nuclear protein from the lower eukaryote Physamm polycephalum munobiology and medicine. Adv. ImmunoI. 33:167-238. homologous to the major proteins of mammalian 40S ribonu- 2. Tan, E.M., and H.G. Kunkel. 1966. Characteristics of a soluble cleoprotein particles. Biochem. Biophys. Res. Commun. 74: nuclear antigen precipitating with sera of patients with sys- 621-629. temic lupus erythematosus. J. Immunol. 96:464-469. 19. Henriquez, K., G. Blobel, and J.P. Aris. 1990. Isolation and 3. Shero, J.H., B. BordweU, N.I. Kothfield, and W.C. Earnshaw. sequencing of NOPI: a yeast gene encoding a nucleolar pro- 1986. High titers of autoantibodies to topoisomerase I (Scl- tein homologous to a human autoimmune antigen. J. Biol. 70) in sera from scleroderma patients. Science (Wash. DC). Chem. 265:2209-2215. 231:737-740. 20. Guiltinan, M.J., M.E. ScheUing, N.Z. Ehtesham, J.C. Thomas, 4. Fritzler, M.J., T.D. Kinsella, and E. Gabult. 1980. The CREST and M.E. Christensen. 1988. The nucleolar RNA binding pro- syndrome: a distinct serologic entity with anticentromere an- tein B-36 is highly conserved among plants. Eur.J. Cell. Biol. tibodies. Am. J. Med. 69:520-526. 46:547-553. 5. Mathews, M.B., and R.M. Bernstein. 1983. Myositis autoan- 21. Aris, J.P., and G. Blobel. 1988. Identification and character- tibody inhibits histidyl-tRNA synthetase: a model for autoim- ization of yeast nucleolar protein that is similar to a rat liver munity. Nature (Lond.). 304:177-179. nucleolar protein. J. Cell Biol. 107:17-31. 6. Pinnas, J.L., J.D. Northway, and E.M. Tan. 1973. Antinucleolar 22. Reimer, G., V.D. Steen, C.A. Penning, T.A. Medsger, and E.M. antibodies in human sera. J. Immunol. 111:996-1004. Tan. 1988. Correlates between autoantibodies to nucleolar an- Downloaded from http://rupress.org/jem/article-pdf/181/3/1027/1106300/1027.pdf by guest on 25 September 2021 7. Reimer, G., K.M. Pose, U. Scheer, and E.M. Tan. 1987. Au- tigens and clinical features in patients with systemic sclerosis. toantibody to KNA polymerase I in scleroderma sera. J. Clin. Arthritis Rheum. 31:525-532. Invest. 79:65-72. 23. Okano, Y., V.D. Steen, and T.A. Medsger. 1992. Autoanti- 8. Keddy, K., E.M. Tan, D. Henning, K. Nogha, and H. Busch. body to U3 nucleolar ribonucleoprotein (fibrillarin) in patients 1983. Detection of a nucleolar 7-2 ribonucleoprotein and cy- with systemic sclerosis. Arthritis Rheum. 35: 95-100. toplasmic 8-2 ribonucleoprotein with autoantibodies from pa- 24. Kuwana, M., J. Kaburaki., Y. Okano., T Tojo, and M. Homma. tients with scleroderma. J. Biol. Chem. 258:1383-1386. 1994. Clinical and prognostic associations based on serum anti- 9. Reimer, G., U. Scheer, I.M. Peters, and E.M. Tan. 1986. Im- nuclear antibodies in Japanese patients with systemic sclerosis. munolocalization and partial characterization of nuclear au- Arthritis Rheum. 37:75-83. toantigen (PM-Scl) associated with polymyositisAcleroderma 25. Parker, K.A., and J.A. Steitz. 1987. Structural analyses of overlap syndrome, j. Immunol. 137:3802-3808. human U3 ribonucleoprotein particle reveal a conserved se- 10. Ochs, R.L., M.A. Lischwe, W.H. Spohn, and H. Busch, 1985. quence available for base pairing with pre-rKNA. Mol. Cell. A new protein of the nucleolus identified by autoimmune sera. Biol. 7:2899-2913. Biol. Cell. 54:123-134. 26. Baserga, S.J., X.W. Yang, andJ.A. Steitz. 1991. An intact box 11. Lischwe, M.A., K.L. Ochs, K. Reddy, K.G. Cook, L.C. Yeo- C sequence in the U3 snRNA is required for binding of fibril- man, E.M. Tan, M. Reichlin, and H. Busch. 1985. Purification latin, the protein common to the major family of nucleolar and partial characterization of a nucleolar scleroderma antigen snRNP. EMBO. (Eur. Mol. Biol. Organ.) J. 10:2645-2651. rich in Ng,Ng-dimethylarginine. J. Biol. Chem. 260:14304- 27. Monestier, M., M.J. Losman., K.E. Novick, andJ.P. Aris. 1994. 14310. Molecular analysis of mercury-induced antinucleolar antibodies 12. Kurzhals, G., M. Meurer, T. Krieg, and G. Keimer. 1990. Clin- in H-2s mice. J. Immunol. 152:667-675. ical association of autoanribodies to fibtillarin with diffuse 28. Smith, D.B., and K.S. Johnson. 1988. Single step purification scleroderma and disseminated telangiectasia.J. Am. Acad. Der- of polypeptides expressed in E. coli as fusions with glutathione- matol. 23:832-836. S-transferase. Gene. 67:31-40. 13. Hultman, P., S. Enestrom, K.M. Pollard, and E.M. Tan. 1989. 29. Saiki, R.K., D.H. Gelfand, S. Stoffel, S.J. Scharf, R. Higuchi, Anti-fibrillarin autoantibodies in mercury treated mice. Clin. G.T. Hora, K.B. Mullis, and H.A. Erlich. 1988. Primer directed Exp. Immunol. 78:470-472. enzymatic amplification of DNA with a thermostable DNA 14. Reuter, K., G. Tessars, H.W. Vohr, E. Gleichmann, and K. polymerase. Science (Wash. DC). 239:487-491. Luhrmann. 1989. Mercuric chloride induces autoantibodies 30. Sanger, F., S. Nicklsen, and A.K. Goulson. 1977. DNA se- against U3 small nuclear ribonucleoprotein in susceptible mice. quencing with chain terminating inhibitor. Proc. Natl. Acad. Proc. Natl. Acad. Sci. USA. 86:237-241. Sci. USA. 74:5463-5467. 15. Reimer, G., K.M. Pollard, and C.A. Penning. 1987. Mono- 31. Muryoi, T., K.N. Kastuti, M.J. Kafina, D.S. Cram, L.C. clonal autoantibodies and some human scleroderma sera target Harrison, T. Sasaki, and C.A. Bona. 1992. Antitopoisomerase a 34 kDa nucleolar protein of U3-ribonucleoprotein particle. I monoclonal autoantibodies from scleroderma patients and Arthritis Rheum. 30:793-800. tight skin mouse interact with similar epitopes. J. Exp. Med. 16. Aris, J.P., and G. Blobel. 1991. cDNA cloning and sequencing 175:1103-1109. of human fibriUarin, a conserved nucleolar protein recognized 32. Laemmli, U.K. 1970. Cleavage of structural proteins during by autoimmune sera. Proc. Natl. Acad. Sci. USA. 8:931-935. the assembly of the head of bacteriophage T4. Nature (Lond.). 17. Lapeyre, B., P. Mariottini, C. Mathieu, P. Ferrer, F. Amaldi, 227:680-685. E Amalric, and M. Caizergues-Ferrer. 1990. Molecular cloning 33. Towbin, H., T. Staehlin, andJ. Gordon. 1979. Electrophoretic of Xenopus fibtillarin, a conserved U3 small nuclear ribonu- transfer of proteins from polyacrylamide gels to nitrocellulose cleoprotein recognized by antisera from humans with autoim- sheets: procedure and some applications. Proc. Natl. Acad. Sci. mune disease. Mol. Cell. Biol. 10:430-434. USA. 76:4350-4354. 18. Christensen, M.E., A.L. Beyer, B. Walker, and W.M. LeStour- 34. Johnson, G.D., and G.M.C. Noggueria Araujo. 1981. A simple geon. 1977. Identification of NG, NG dimethylarginine in a method of reducing the fading ofimmunofluorescence during

1035 Katsuriet al. microscopy. J. Immunol. Methods. 43:349-350. 41. Cunningham, M.W., N.K. Hall, and K.K. Krisher. 1986. A 35. Imai, H., K.L. Ochs, K. Kiyosawa,S. Furuta, R.M. Nakamura, study of anti-group A streptococcal monoclonal antibodies and E.M. Tan. 1992. Am. J. Pathol. 140:859-866. crossreactive with myosin. J. Immunol. 136:293-298. 36. Arrand, J.K., J.E. Walsh, L. Kymo, E. Bjoerck, T. Lindahl, 42. Kagnoff, M.F., R.K. Austin, J.J. Hubert, J.E. Bernardin, and and B.E. Griffin. 1981. Molecular cloning of the complete D.D. Kasarda. 1984. Possible role of a human adenovirus in Epstein-Barr virus genome as set of overlapping restriction en- the pathogenesis of celiac disease.J. Extx Med. 160:1544-1557. donuclease fragments. Nucleic Acids. Res. 9:2999-3014. 43. Nickowitz, R.E., and H. Worman. 1993. Autoantibodieswith 37. Yanagisawa, M., Y. Hamada, Y. Katsuragawa, M. Imamura, primary biliary cirrhosis recognize a restricted region within T. Mikawa, and T. Masaki. 1987. Complete primary structure the cytoplasmic tail of nuclear pore membrane glycoprotein of vertebrate smooth muscle myosin heavychain deduced from Gp210. J. Exp. Med. 178:2237-2242. its complementaryDNA sequence. Implication on topography 44. Reeves, W.H., and Z.M. Sthoeger. 1989. Molecular cloning and function of myosin. J. Mol. Biol. 198:143-157. of cDNA encoding the P70 (Ku) lupus autoantigen. J. Biol. 38. McGeoch, D.J., M.A. Dalrymple, A.J. Davison, A. Dolan, Chem. 264:5047-5052. M.C. Frame, D. McNab, L.J. Perry, J.E. Scott, and P. Taylor. 45. Rockeach,L.A.,J.A. Haselby,J.E Meilof, R.J.T. Smeenk, T.R. 1988. The complete DNA sequence of the long unique region Unnesch, R.M. Greene, and S.D. Hoch. 1991. Characteriza- in the genome of herpes simplex virus type 1. J. Gen. Virol. tion of the autoantigen calreticulin.J. Immunol. 147:3031-3039. 69:1531-1574. 46. Maul, G.G., S.A. Jimenez, E. Riggs, and D. Ziemnicka-Kotula.

39. Fujinami, R.S., and M.B.A. Oldstone. 1985. Amino acid ho- 1989. Determination of an epitope of diffusesystemic sclerosis Downloaded from http://rupress.org/jem/article-pdf/181/3/1027/1106300/1027.pdf by guest on 25 September 2021 mology between the encephalitogenicsite of myelin basic pro- marker antigen, DNA topoisomeraseI: sequencesimilarity with tein and virus: mechanism for autoimmunity. Science (Wash. retroviral P30gag protein suggests a possible cause for autoim- DC). 230:1043-1045. munity in systemic sclerosis. Proc. Natl. Acad. Sci. USA. 40. Nickerson,C., H. Luttua, and C. David. 1991. Antigenic mim- 86:8492-8496. icry and autoimmune diseases. Int. Rw. Immunol. 7:205-225.

1036 FibrillarinEpitopes Recognizedby Autoantibodies